Conference Paper
BibTex RIS Cite

Akciğer Kanseri Tespitinde Sınıf Aktivasyon Haritaları Kullanarak Açıklanabilir Derin Öğrenme Modeli ve Radyolog Değerlendirmesi

Year 2024, Volume: 04 Issue: 02, 166 - 175, 31.12.2024

Abstract

Akciğer kanseri dünya genelinde yaygın ve ölümcül olan kanser türlerinden biridir. Akciğer kanserinin erken tanısı, hastalığın tedavi edilebilir aşamada tespit edilmesine olanak tanır ve bu da hastanın yaşam şansını önemli ölçüde artırır. Yapay zekânın bu alanda kullanımı, bir dizi avantaj sağlayarak hastalığın daha etkili bir şekilde tespitine yardımcı olabilir. Son zamanlarda derin öğrenme yaklaşımları görüntü sınıflandırma çalışmalarında baskın bir rol almaktadır. Derin öğrenme yöntemlerinin en önemli dezavantajlarından birisi kapalı-kutu yapısı nedeniyle güvenirlik açısından şeffaflık eksikliğidir. Bu amaçla açıklanabilir modeller önemli bir araştırma haline gelmiştir. Açıklanabilir derin öğrenme modelleri, genellikle sınıf aktivasyon haritaları (Class Activation Maps - CAM) gibi tekniklere dayanmaktadır. Bu çalışmada, akciğer kanseri tespitinde açıklanabilir bir derin öğrenme modeli oluşturulmuştur. Makale kapsamındaki deneysel çalışmalar, akciğer bilgisayarlı tomografi (BT) görüntülerini içeren ve açık erişimli bir veri seti üzerinde yürütülmüştür. Sınıflandırma aşamasında konvolüsyonel sinir ağları (KSA) tabanlı ResNet101V2, VGG16, MobileNetV2, DenseNet201 ve EfficientNetB0 modelleri kullanılmıştır. Deneysel sonuçlara göre EfficientNetB0 modeli %98.63 ile en yüksek sınıflandırma doğruluğuna ulaşmıştır. Sınıflandırıcı modele uygulanan farklı CAM teknikleri ile ısı haritaları elde edilmiştir. Elde edilen ısı haritaları uzman radyolog tarafından değerlendirilerek sonuçlar tartışılmıştır.

References

  • [1] Witschi H. A short history of lung cancer. Toxicol Sci. 2001;64(1):4–6.
  • [2] Kaya U, Yılmaz A, Dikmen Y, Beigelman-Aubry C, Dunet V, Brun AL, et al. Comparison of the automatic segmentation of multiple organs at risk in CT images of lung cancer between deep convolutional neural network-based and atlas-based techniques. Diagn Interv Imaging [Internet]. 2019;97(16):973–89. Available from: https://doi.org/10.1080/0284186X.2018.1529421
  • [3] Espinoza JL, Dong LT. Artificial intelligence tools for refining lung cancer screening. J Clin Med. 2020;9(12):1–17.
  • [4] Christie JR, Lang P, Zelko LM, Palma DA, Abdelrazek M, Mattonen SA. Artificial Intelligence in Lung Cancer: Bridging the Gap Between Computational Power and Clinical Decision-Making. Can Assoc Radiol J. 2021;72(1):86–97.
  • [5] Radhika PR, Nair RAS, Veena G. A Comparative Study of Lung Cancer Detection using Machine Learning Algorithms. Proc 2019 3rd IEEE Int Conf Electr Comput Commun Technol ICECCT 2019. 2019;1–4.
  • [6] Ait Skourt B, El Hassani A, Majda A. Lung CT image segmentation using deep neural networks. Procedia Comput Sci [Internet]. 2018;127:109–13. Available from: https://doi.org/10.1016/j.procs.2018.01.104
  • [7]. Yokota K, Maeda S, Kim H, Tan JK, Ishikawa S, Tachibana R, et al. Automatic detection of GGO regions on CT images in LIDC dataset based on statistical features. 2014 Jt 7th Int Conf Soft Comput Intell Syst SCIS 2014 15th Int Symp Adv Intell Syst ISIS 2014. 2014;1374–7.
  • [8] Mhaske D, Rajeswari K, Tekade R. Deep learning algorithm for classification and prediction of lung cancer using CT scan images. Proc - 2019 5th Int Conf Comput Commun Control Autom ICCUBEA 2019. 2019;
  • [9] Abdul W. An Automatic Lung Cancer Detection and Classification (ALCDC) System Using Convolutional Neural Network. Proc - Int Conf Dev eSystems Eng DeSE. 2020;2020-Decem:443–6.
  • [10] Chen Y, Wang Y, Hu F, Feng L, Zhou T, Zheng C. Ldnnet: Towards robust classification of lung nodule and cancer using lung dense neural network. IEEE Access. 2021;9:50301–20.
  • [11] Bharati S, Podder P, Paul PK. Lung cancer recognition and prediction according to random forest ensemble and RUSBoost algorithm using LIDC data. Int J Hybrid Intell Syst. 2019;15(2):91–100.
  • [12] Chen L, Chen J, Hajimirsadeghi H, Mori G. Adapting grad-CAM for embedding networks. Proc - 2020 IEEE Winter Conf Appl Comput Vision, WACV 2020. 2020;2783–92.
  • [13] Selvaraju RR, Cogswell M, Das A, Vedantam R, Parikh D, Batra D. Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization. Proc IEEE Int Conf Comput Vis. 2017;2017-Octob:618–26.
  • [14] Panwar H, Gupta PK, Siddiqui MK, Morales-Menendez R, Bhardwaj P, Singh V. A deep learning and grad-CAM based color visualization approach for fast detection of COVID-19 cases using chest X-ray and CT-Scan images. Chaos, Solitons and Fractals [Internet]. 2020;140:110190. Available from: https://doi.org/10.1016/j.chaos.2020.110190
  • [15] Pradhan K, Chawla P. Medical Internet of things using machine learning algorithms for lung cancer detection. J Manag Anal. 2020;7(4):591–623.
  • [16] Zakriya KJ, Christmas C, Wenz JF, Franckowiak S, Anderson R, Sieber FE. Preoperative Factors Associated with Postoperative Change. 2002;1628–32.
  • [17] Fan BB, Yang H. Analysis of identifying COVID-19 with deep learning model. J Phys Conf Ser. 2020;1601(5).
  • [18] alyasriy hamdalla, AL-Huseiny M. The IQ-OTH/NCCD lung cancer dataset. 2023;4.
  • [19] Goodfellow, I., Bengio, Y., & Courville A. Deep learning. 2016.
  • [20] Gong W, Chen H, Zhang Z, Zhang M, Gao H. A Data-Driven-Based Fault Diagnosis Approach for Electrical Power DC-DC Inverter by Using Modified Convolutional Neural Network with Global Average Pooling and 2-D Feature Image. IEEE Access. 2020;8:73677–97.

Explainable Deep Learning Model Using Class Activation Maps in Lung Cancer Testing and Radiologist Evaluation

Year 2024, Volume: 04 Issue: 02, 166 - 175, 31.12.2024

Abstract

Lung cancer is one of the most common and deadly cancer types worldwide. Early diagnosis of lung cancer allows detecting the disease at a treatable stage, which significantly increases the patient's chances of survival. The use of artificial intelligence in this field can help detect the disease more effectively, providing a number of advantages. Recently, deep learning approaches have taken a dominant role in image classification studies. One of the most important disadvantages of deep learning methods is the lack of transparency in terms of reliability due to their closed-box structure. For this purpose, explainable models have become an important research area. Explainable deep learning models are generally based on techniques such as class activation maps (CAM). In this study, an explainable deep learning model was created for lung cancer detection. The experimental studies within the scope of the article were conducted on an open-access dataset containing lung computed tomography (CT) images. In the classification phase, convolutional neural networks (CNN) based ResNet101V2, VGG16, MobileNetV2, DenseNet201 and EfficientNetB0 models were used. According to experimental results, the EfficientNetB0 model reached the highest classification accuracy of 98.63%. Heat maps were obtained with different CAM techniques applied to the classifier model. The resulting heat maps were evaluated by the expert radiologist and the results were discussed.

References

  • [1] Witschi H. A short history of lung cancer. Toxicol Sci. 2001;64(1):4–6.
  • [2] Kaya U, Yılmaz A, Dikmen Y, Beigelman-Aubry C, Dunet V, Brun AL, et al. Comparison of the automatic segmentation of multiple organs at risk in CT images of lung cancer between deep convolutional neural network-based and atlas-based techniques. Diagn Interv Imaging [Internet]. 2019;97(16):973–89. Available from: https://doi.org/10.1080/0284186X.2018.1529421
  • [3] Espinoza JL, Dong LT. Artificial intelligence tools for refining lung cancer screening. J Clin Med. 2020;9(12):1–17.
  • [4] Christie JR, Lang P, Zelko LM, Palma DA, Abdelrazek M, Mattonen SA. Artificial Intelligence in Lung Cancer: Bridging the Gap Between Computational Power and Clinical Decision-Making. Can Assoc Radiol J. 2021;72(1):86–97.
  • [5] Radhika PR, Nair RAS, Veena G. A Comparative Study of Lung Cancer Detection using Machine Learning Algorithms. Proc 2019 3rd IEEE Int Conf Electr Comput Commun Technol ICECCT 2019. 2019;1–4.
  • [6] Ait Skourt B, El Hassani A, Majda A. Lung CT image segmentation using deep neural networks. Procedia Comput Sci [Internet]. 2018;127:109–13. Available from: https://doi.org/10.1016/j.procs.2018.01.104
  • [7]. Yokota K, Maeda S, Kim H, Tan JK, Ishikawa S, Tachibana R, et al. Automatic detection of GGO regions on CT images in LIDC dataset based on statistical features. 2014 Jt 7th Int Conf Soft Comput Intell Syst SCIS 2014 15th Int Symp Adv Intell Syst ISIS 2014. 2014;1374–7.
  • [8] Mhaske D, Rajeswari K, Tekade R. Deep learning algorithm for classification and prediction of lung cancer using CT scan images. Proc - 2019 5th Int Conf Comput Commun Control Autom ICCUBEA 2019. 2019;
  • [9] Abdul W. An Automatic Lung Cancer Detection and Classification (ALCDC) System Using Convolutional Neural Network. Proc - Int Conf Dev eSystems Eng DeSE. 2020;2020-Decem:443–6.
  • [10] Chen Y, Wang Y, Hu F, Feng L, Zhou T, Zheng C. Ldnnet: Towards robust classification of lung nodule and cancer using lung dense neural network. IEEE Access. 2021;9:50301–20.
  • [11] Bharati S, Podder P, Paul PK. Lung cancer recognition and prediction according to random forest ensemble and RUSBoost algorithm using LIDC data. Int J Hybrid Intell Syst. 2019;15(2):91–100.
  • [12] Chen L, Chen J, Hajimirsadeghi H, Mori G. Adapting grad-CAM for embedding networks. Proc - 2020 IEEE Winter Conf Appl Comput Vision, WACV 2020. 2020;2783–92.
  • [13] Selvaraju RR, Cogswell M, Das A, Vedantam R, Parikh D, Batra D. Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization. Proc IEEE Int Conf Comput Vis. 2017;2017-Octob:618–26.
  • [14] Panwar H, Gupta PK, Siddiqui MK, Morales-Menendez R, Bhardwaj P, Singh V. A deep learning and grad-CAM based color visualization approach for fast detection of COVID-19 cases using chest X-ray and CT-Scan images. Chaos, Solitons and Fractals [Internet]. 2020;140:110190. Available from: https://doi.org/10.1016/j.chaos.2020.110190
  • [15] Pradhan K, Chawla P. Medical Internet of things using machine learning algorithms for lung cancer detection. J Manag Anal. 2020;7(4):591–623.
  • [16] Zakriya KJ, Christmas C, Wenz JF, Franckowiak S, Anderson R, Sieber FE. Preoperative Factors Associated with Postoperative Change. 2002;1628–32.
  • [17] Fan BB, Yang H. Analysis of identifying COVID-19 with deep learning model. J Phys Conf Ser. 2020;1601(5).
  • [18] alyasriy hamdalla, AL-Huseiny M. The IQ-OTH/NCCD lung cancer dataset. 2023;4.
  • [19] Goodfellow, I., Bengio, Y., & Courville A. Deep learning. 2016.
  • [20] Gong W, Chen H, Zhang Z, Zhang M, Gao H. A Data-Driven-Based Fault Diagnosis Approach for Electrical Power DC-DC Inverter by Using Modified Convolutional Neural Network with Global Average Pooling and 2-D Feature Image. IEEE Access. 2020;8:73677–97.
There are 20 citations in total.

Details

Primary Language Turkish
Subjects Software Engineering (Other)
Journal Section Research Article
Authors

Savaş Tunçer 0000-0003-2455-5388

Oğuzhan Katar 0000-0002-5628-3543

Tülin Öztürk 0000-0001-8942-5264

Özal Yıldırım 0000-0001-5375-3012

Publication Date December 31, 2024
Submission Date July 24, 2024
Acceptance Date October 16, 2024
Published in Issue Year 2024 Volume: 04 Issue: 02

Cite

IEEE S. Tunçer, O. Katar, T. Öztürk, and Ö. Yıldırım, “Akciğer Kanseri Tespitinde Sınıf Aktivasyon Haritaları Kullanarak Açıklanabilir Derin Öğrenme Modeli ve Radyolog Değerlendirmesi”, Researcher, vol. 04, no. 02, pp. 166–175, 2024.

The journal "Researcher: Social Sciences Studies" (RSSS), which started its publication life in 2013, continues its activities under the name of "Researcher" as of August 2020, under Ankara Bilim University.
It is an internationally indexed, nationally refereed, scientific and electronic journal that publishes original research articles aiming to contribute to the fields of Engineering and Science in 2021 and beyond.
The journal is published twice a year, except for special issues.
Candidate articles submitted for publication in the journal can be written in Turkish and English. Articles submitted to the journal must not have been previously published in another journal or sent to another journal for publication.